Artificial Intelligence Risk Evaluation Act of 2025 (S. 2938)

Summary

Senate bill that would require the Secretary of Energy to set up an Advanced Artificial Intelligence Evaluation Program within 90 days of enactment. Developers of advanced AI systems, trained with more than 10^26 operations, would have to take part and provide code, training data, model weights and other details on request, and nobody could deploy such a system without complying, on pain of fines of at least $1,000,000 a day. The program would test for adverse AI incidents such as loss of control, weaponisation and scheming behaviour, and DOE would recommend a permanent federal oversight framework, including possible licensing.

Safety topics

Status notes

Introduced on 29 September 2025 by Sen. Josh Hawley (R-MO), with Sen. Richard Blumenthal (D-CT) as cosponsor, and referred to the Senate Committee on Commerce, Science, and Transportation. Sen. Marsha Blackburn (R-TN) joined as a cosponsor on 19 March 2026. The latest action is a hearing held by the Senate Committee on Homeland Security and Governmental Affairs on 30 September 2026. DOE's recommendation for a permanent oversight framework would be due within 360 days of enactment and updated every year; the program would also develop options for preventing or managing the development of artificial superintelligence. The program would end seven years after enactment unless Congress renews it. The Secretary could redefine 'advanced artificial intelligence system' only by a rule that takes effect once a joint resolution approving it is enacted.

Sources

Open the official source https://www.govinfo.gov/app/details/BILLS-119s2938is

Secondary source: https://www.govinfo.gov/content/pkg/BILLS-119s2938is/html/BILLS-119s2938is.htm

Local copy

Verification

History

1. 2026-10-18notes: Introduced 29 Sep 2025 and referred to the Senate Commerce, Science, and Transpo → Introduced on 29 September 2025 by Sen. Josh Hawley (R-MO), with Sen. Richard Bl — rewritten from the bill and the bill status file: sponsors with parties, the new cosponsor, the latest action (hearing of 30 Sep 2026), deadlines, sunset and the redefinition rule; the remark about what could not be verified moved to verification.note2026-10-17-recheck-north-america-1 · Claude Code (claude-opus-5-5)
2. 2026-10-18safety_topics: ["general-purpose / frontier AI", "evaluation & testing", "catastrophic risk", " → ["general-purpose / frontier AI", "evaluation & testing", "catastrophic risk", " — 'incident reporting' removed: the bill evaluates the likelihood of adverse AI incidents but imposes no incident reporting2026-10-17-recheck-north-america-1 · Claude Code (claude-opus-5-5)
3. 2026-10-18summary: Bipartisan bill that would create a mandatory Department of Energy 'Advanced Art → Senate bill that would require the Secretary of Energy to set up an Advanced Art — 'Bipartisan' removed (not in the stored texts); deployment ban, the 90-day start and the superintelligence options stated as the bill words them2026-10-17-recheck-north-america-1 · Claude Code (claude-opus-5-5)

Related records

FRONTIER Act United States · Draft / proposed

Citation

U.S. Senate (Sen. Josh Hawley, Sen. Richard Blumenthal). Artificial Intelligence Risk Evaluation Act of 2025 (S. 2938) [S. 2938, 119th Congress]. 2025-09-29. https://www.govinfo.gov/app/details/BILLS-119s2938is. Status: Draft / proposed; verified 2026-10-18 (level A). Local copy archive/US-18/2026-10-11/official.html retrieved 2026-10-11 14:03:59 UTC, SHA-256 f214febf5a6c57e9123e7b46c7e2294a9d7bb3dbec8efb06b6492b0033da75e3. Every AI Policy Ever, version 1.18 (snapshot 2026-10-18), record US-18.

Improve this record

Know of an amendment, a status change or an error? Report news or an error on GitHub; the form opens with this record filled in.